Removal of Personal Data from Ticket Examples

Tip

Enneo uses past customer correspondence to generate better response suggestions for new customer service inquiries.

Since customer correspondence typically contains personal data within the meaning of Art. 4 No. 1 GDPR, Enneo removes this information in a five-step process before using examples to improve response suggestions.

1. Overview of the process

Initially, we apply an advanced Language Model (LLM), a form of sophisticated artificial intelligence. This model is specifically trained to analyze large volumes of text efficiently. It reliably recognizes patterns indicating personal information in customer emails, like names or contact details. Through continuous updates and improvements, we keep the LLM up-to-date in dealing with personal data.

2. Recognition of personal data

The LLM thoroughly scrutinizes each email and identifies personal data such as names, addresses, phone numbers, and email addresses. It uses advanced algorithms to detect structures and patterns typical for such information.

3. Replacement with placeholders

After identification, personal details are replaced with neutral placeholders like "xxxx xxxx". This step is carried out in two stages (deterministic pre-processing, subsequently AI-assisted detection) to increase the detection rate. The generated examples no longer contain any direct personal details, and the residual risk that individual details remain undetected is minimized by the two-step process and monitored by random sample checks (Step 4).

4. User interface and manual checks

A specially developed user interface provides controlled access to the cleaned data. This interface is intuitively designed and allows authorized employees to manage and monitor data efficiently. Regular random sample checks are carried out. These checks serve to ensure the quality of the procedure; if necessary, corrections are made.

5. Data protection and security

The entire process is designed for data protection and security. The removal of personal data before using it for response suggestions is a central part of our commitment to our customers' privacy. Access to both cleaned and uncleaned data is limited to authorized employees. The process as well as any remaining risks are continually assessed in the Data Protection Impact Assessment (DPIA).